I ran a certbot certbot renew --dry-run and there are a couple of dead domains, but that's expected. Is this just snap's way of saying some domains failed to renew? As far as I can tell everything else is OK.
That is normal. Let's Encrypt certs no longer use OCSP (use CRL instead). Certbot supports other Certificate Authorities so the message might apply to those.
I should have had you check the service too.
sudo systemctl status --no-pager -l snap.certbot.renew.service
sudo journalctl -u certbot.renew
I am hoping to get more details on why it did not start when it apparently did.
Perhaps Certbot exits with a non-zero exit code because of your "dead domains". And, as you guessed, maybe that is the cause of this start failure log message. You had this kind of failure before although we saw messages describing the non-zero exit code: A question about certbot apt vs snap on Ubuntu - #6 by gilgongo
I am not expert with snap debugging these are just routine systemd checks. If no one else here has suggestions you might try the EFF's github for Certbot (here). Or, maybe even the Snapcraft support forum: https://forum.snapcraft.io/
Well, the error about "Failed to start" is different than failed with exit-code as it did before.
So, there may be something different this time. Really not sure. You could cleanup your failing renewals and see if it continues. You should do that anyway
If you don't need certs use:
sudo certbot delete --cert-name X
Where X is the cert-name from: sudo certbot certificates
Just make sure no service references the cert files otherwise those may fail (Apache, mail systems, ...)